To specify the location of perceptually dominant markers for temporal structures of speech, the acceptability or detectability of the modification of segmental duration is measured. The modification is carried out in a complementary way, i.e., two successive segments are lengthened or shortened, and have the same absolute duration and opposite directions of change. The first experiment using 15 four-mora word stimuli shows that a vowel (V) duration and its adjacent consonant (C) duration can perceptually compensate each other. This compensation is found to not depend on the temporal order of target pairs (C-to-V or V-to-C), but rather on the loudness difference between V and C; the acceptability decreases when the loudness difference between V and C becomes high. This suggests that perceptually dominant markers locate around major loudness jumps; this finding is supported by the second experiment using non-speech stimuli replicating the loudness contours of speech stimuli. The detectability of temporal displacement is higher at large loudness jumps than at small loudness jumps.